Some interviews take an identity capture before starting: a photo, a spoken pledge, or both. It is shown as a step with an explanation; it is not taken silently.
The reference photo
A still from your camera, used as the reference for face checks during the interview. Look at the camera, keep your face lit from the front, and remove anything covering it.
It exists so that "is this the same person throughout" can be answered at all. Without a reference there is nothing to compare later frames against.
The spoken pledge
A short sentence read aloud, confirming the interview is your own work. It is transcribed and checked against what you were asked to say.
Two things about it are worth knowing. It is in the interview's language, so a Tamil interview asks for a Tamil pledge. And it contains a couple of words specific to your sitting — that is what makes it evidence of this interview rather than a clip that could have been recorded once and reused.
If it fails
Retry. The usual causes are ordinary: poor lighting, a microphone that was not picking up, speaking too quietly or too fast.
Failing repeatedly does not end your interview by itself. The attempts are recorded and the interview continues, with the identity check noted as unverified rather than as misconduct — a failed camera and a false identity are different things and are recorded differently.
Opting out
Where an organisation has made identity capture optional you will see a way to decline, and declining is recorded rather than penalised. Where it is required, it is required, and the alternative is to raise it with the organisation.